Horton

Re: Removing unneeded self joins

Andrei Lepikhov <a.lepikhov@postgrespro.ru>

View thread
From: Andrei Lepikhov <a.lepikhov@postgrespro.ru>
To: Alexander Korotkov <aekorotkov@gmail.com>
Cc: "Gregory Stark (as CFM)" <stark.cfm@gmail.com>, Michał Kłeczek <michal@kleczek.org>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-10-05T09:17:08Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Remove GUC_NOT_IN_SAMPLE from enable_self_join_elimination

  2. Put enable_self_join_elimination into postgresql.conf.sample

  3. Get rid of ojrelid local variable in remove_rel_from_query()

  4. Implement Self-Join Elimination

  5. Revert: Remove useless self-joins

  6. Replace lateral references to removed rels in subqueries

  7. Replace relids in lateral subquery parse tree during SJE

  8. Forbid SJE with result relation

  9. Fix misuse of RelOptInfo.unique_for_rels cache by SJE

  10. Replace the relid in some missing fields during SJE

  11. Revert 56-bit relfilenode change and follow-up commits.

  12. Stabilize timetz test across DST transitions.

  13. Speed up finding EquivalenceClasses for a given set of rels

  14. Fix mark-and-restore-skipping test case to not be a self-join.

Attachments

On 4/10/2023 14:34, Alexander Korotkov wrote:
>  > Relid replacement machinery is the most contradictory code here. We used
>  > a utilitarian approach and implemented a simplistic variant.
> 
>  > > 2) It would be nice to skip the insertion of IS NOT NULL checks when
>  > > they are not necessary.  [1] points that infrastructure from [2] might
>  > > be useful.  The patchset from [2] seems committed mow.  However, I
>  > > can't see it is directly helpful in this matter.  Could we just skip
>  > > adding IS NOT NULL clause for the columns, that have
>  > > pg_attribute.attnotnull set?
>  > Thanks for the links, I will look into that case.
To be more precise, in the attachment, you can find a diff to the main 
patch, which shows the volume of changes to achieve the desired behaviour.
Some explains in regression tests shifted. So, I've made additional tests:

DROP TABLE test CASCADE;
CREATE TABLE test (a int, b int not null);
CREATE UNIQUE INDEX abc ON test(b);
explain SELECT * FROM test t1 JOIN test t2 ON (t1.a=t2.a)
WHERE t1.b=t2.b;
CREATE UNIQUE INDEX abc1 ON test(a,b);
explain SELECT * FROM test t1 JOIN test t2 ON (t1.a=t2.a)
WHERE t1.b=t2.b;
explain SELECT * FROM test t1 JOIN test t2 ON (t1.a=t2.a)
WHERE t1.b=t2.b AND (t1.a=t2.a OR t2.a=t1.a);
DROP INDEX abc1;
explain SELECT * FROM test t1 JOIN test t2 ON (t1.a=t2.a)
WHERE t1.b=t2.b AND (t1.b=t2.b OR t2.b=t1.b);

We have almost the results we wanted to have. But in the last explain 
you can see that nothing happened with the OR clause. We should use the 
expression mutator instead of walker to handle such clauses. But It 
doesn't process the RestrictInfo node ... I'm inclined to put a solution 
of this issue off for a while.

-- 
regards,
Andrey Lepikhov
Postgres Professional